mirror of
https://github.com/ollama/ollama.git
synced 2026-01-12 00:06:57 +08:00
33ee7168ba
* WIP - MLX backend with gemma3 * MLX: add cmake and go tag build toggles To build the new MLX backend code: cmake --preset MLX cmake --build --preset MLX --parallel cmake --install build --component MLX go build -tags mlx . Note: the main.go entrypoint for the MLX engine will change in a follow up commit. * add experimental image generation runtime * add experimental image generation runtime * MLX: wire up cuda build for linux * MLX: get dependencies correct and dedup This is still too large for a unified github artifact, but is now "correct" for the mlx_cuda_v13 directory. * fix relative link bug in dedup * Add darwin build and readme * add go build tag for mlx dependent code and wire up build_darwin.sh * lint cleanup * macos: build mlx for x86 This will be CPU only. * cuda build instructions and fix drift from mlx bump * stale comment * Delete agent helper doc * Clean up readme.md * Revise README for tokenizer clarity and details Updated README to clarify tokenizer functionality and removed correctness section. --------- Co-authored-by: jmorganca <jmorganca@gmail.com>
67 lines
1.5 KiB
Go
67 lines
1.5 KiB
Go
//go:build mlx
|
|
|
|
package qwen_image
|
|
|
|
import (
|
|
"os"
|
|
"testing"
|
|
|
|
"github.com/ollama/ollama/x/imagegen/mlx"
|
|
)
|
|
|
|
// TestPipelineOutput runs the full pipeline (integration test).
|
|
// Skips if model weights not found. Requires ~50GB VRAM.
|
|
func TestPipelineOutput(t *testing.T) {
|
|
modelPath := "../../../weights/Qwen-Image-2512"
|
|
if _, err := os.Stat(modelPath); os.IsNotExist(err) {
|
|
t.Skip("Skipping: model weights not found at " + modelPath)
|
|
}
|
|
|
|
// Load model
|
|
pm, err := LoadPersistent(modelPath)
|
|
if err != nil {
|
|
t.Skipf("Skipping: failed to load model: %v", err)
|
|
}
|
|
|
|
// Run 2-step pipeline (minimum for stable scheduler)
|
|
cfg := &GenerateConfig{
|
|
Prompt: "a cat",
|
|
Width: 256,
|
|
Height: 256,
|
|
Steps: 2,
|
|
Seed: 42,
|
|
}
|
|
|
|
output, err := pm.GenerateFromConfig(cfg)
|
|
if err != nil {
|
|
t.Fatalf("Pipeline failed: %v", err)
|
|
}
|
|
mlx.Eval(output)
|
|
|
|
// Verify output shape [1, C, H, W]
|
|
shape := output.Shape()
|
|
if len(shape) != 4 {
|
|
t.Errorf("Expected 4D output, got %v", shape)
|
|
}
|
|
if shape[0] != 1 || shape[1] != 3 || shape[2] != cfg.Height || shape[3] != cfg.Width {
|
|
t.Errorf("Shape mismatch: got %v, expected [1, 3, %d, %d]", shape, cfg.Height, cfg.Width)
|
|
}
|
|
|
|
// Verify values in expected range [0, 1]
|
|
data := output.Data()
|
|
minVal, maxVal := float32(1.0), float32(0.0)
|
|
for _, v := range data {
|
|
if v < minVal {
|
|
minVal = v
|
|
}
|
|
if v > maxVal {
|
|
maxVal = v
|
|
}
|
|
}
|
|
t.Logf("Output range: [%.4f, %.4f]", minVal, maxVal)
|
|
|
|
if minVal < -0.1 || maxVal > 1.1 {
|
|
t.Errorf("Output values out of range: [%.4f, %.4f]", minVal, maxVal)
|
|
}
|
|
}
|